A high-performance CPU, GPU, and memory profiler for Python with AI-powered optimization suggestions.
Scalene is a high-performance profiler for Python that analyzes CPU, GPU, and memory usage with exceptional detail and speed. It helps developers identify performance bottlenecks and memory issues in their code, and uniquely offers AI-powered suggestions for optimization.
Python developers and data scientists who need to optimize application performance, debug memory leaks, or reduce computational costs in production code, research scripts, or data pipelines.
Developers choose Scalene for its combination of low overhead, line-level precision across CPU/GPU/memory, and integrated AI optimization proposals, which together provide a more actionable and comprehensive profiling experience than traditional tools.
Scalene: a high-performance, high-precision CPU, GPU, and memory profiler for Python with AI-powered optimization proposals
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Connects to multiple AI providers like OpenAI, Azure, and Ollama to generate code improvement suggestions directly in the profiling interface, making optimizations actionable.
Uses efficient sampling with typically 10-20% overhead, running orders of magnitude faster than many profilers while maintaining high accuracy.
Provides detailed CPU, memory, and GPU profiles per line of code, separating Python, native, and system time for precise bottleneck identification.
Automatically identifies lines responsible for likely memory leaks and reports copying volume, aiding in debugging memory-intensive applications.
GPU profiling is exclusively available for NVIDIA-based systems, excluding users with AMD or integrated GPUs from this feature.
AI-powered suggestions require API keys and network calls to external services, which can introduce costs, latency, and privacy concerns.
The README notes issues with newer PyTorch versions on macOS, and Windows users need Visual C++ Redistributable, adding setup complexity.